Despite the availability of effective antibiotics and support, infectious diseases continue to cause an unacceptably high morbidity and mortality among elderly, critically ill, and hospitalized patients. Recent animal and human studies imply a significant gender-related influence on both susceptibility to and mortality from infection in both experimental and clinical settings. All clinical studies in this regard, however, have been either small or based on retrospective analysis of prospectively collected data. One explanation for these possible differences in outcome that is supported by animal studies is that elevated circulating levels of estradiol (seen in pre-menopausal adult females, particularly in the peri- ovulatory period) and low systemic testosterone levels are immunologically "protective," mediated by changes in cytokine response. An alternative possibility, however, is related to differences in social/family values that are determinants of patient death, including mode of death (withdrawal of support vs. irreversible physiologic collapse) and assumptions regarding the likelihood of long- term, high quality of life survival. The hypothesis to be tested in this study is that after controlling for confounding variables, high systemic estrogen and low systemic testosterone levels (largely in pre-menopausal women) are associated with a decreased incidence of nosocomial infection, yet the mortality per infection is higher for women (particularly older women) and is dependent on both hormonal and social variables. Aim 1. To prospectively determine the incidence of and associated mortality from nosocomial infections in a critically ill population after controlling for multiple pre-defined confounding variables, including gender. Aim 2. To determine the relationship between sex hormonal status, systemic cytokine levels, and the incidence of and outcome from nosocomial infections, as well as the relative contribution of infection and end of life decisions to outcome in patients dying under study. It is hoped that data obtained from the proposed research can be used to design interventional studies utilizing hormonal manipulation to improve outcomes from otherwise lethal infectious diseases.